Brain-Derived Neurotrophic Factor Plasma Levels: Relationship With Dementia and Diabetes in the Elderly Population

(1)

294

Brain-Derived Neurotrophic Factor Plasma Levels:

Relationship With Dementia and Diabetes in the Elderly

Population

Angela Passaro,

1

_{Edoardo Dalla Nora,}

1

_{Mario L. Morieri,}

1

_{Cecilia Soavi,}

1

_{Juana M. Sanz,}

1

_{Amedeo Zurlo,}

2

Renato Fellin,

1

_{and Giovanni Zuliani,}

1

1_{Department of Clinical and Experimental Medicine, Section of Internal Medicine, Gerontology and Clinical Nutrition,}

University of Ferrara, Italy.

2_{Operative Unit of Geriatrics, Arcispedale S. Anna, Ferrara, Italy.}

Address correspondence to Giovanni Zuliani, PhD, MD, Department of Clinical and Experimental Medicine, Section of Internal Medicine, Gerontology, and Clinical Nutrition, University of Ferrara, Via Savonarola 9, 44100 Ferrara, Italy. Email: [email protected]

The mechanisms linking diabetes and cognitive impairment/dementia, two common conditions of elderly people, are not completely known. Brain-derived neurotrophic factor (BDNF) has antidiabetic properties, and reduced circulat-ing BDNF was associated with dementia. We investigated the relationship between plasma BDNF levels, dementia, and diabetes in a sample of 164 community-dwelling elderly individuals, including 50 participants with vascular demen-tia, 44 with late onset Alzheimer’s disease, 23 with cerebrovascular disease not demendemen-tia, and 47 controls (C). Presence/ absence of diabetes was registered; new diagnoses of diabetes were made by the American Diabetes Association criteria. BDNF plasma levels were measured by ELISA. Both diagnosis of dementia and diabetes were associated with lower BDNF plasma values compared with the respective controls; moreover, dementia and diabetes correlated with BDNF plasma lev-els, independent of possible confounders. A progressive reductions of BDNF plasma levels from C (383.9 ± 204.6 pg/mL), to cerebrovascular disease not dementia (377.1 ± 130.2), to vascular dementia (313.3 ± 114.8), to late onset Alzheimer’s disease (264.7 ± 147.7) was observed, (late onset Alzheimer’s disease vs C, p: .03; late onset Alzheimer’s disease vs cerebrovascular disease not dementia, p: .002). Demented patients affected by diabetes had the lowest BDNF mean levels (264.9 pg/mL) among individuals enrolled in this sample, suggesting the existence of a “synergistic” effect of dementia and diabetes on BDNF levels.

Key Words: BDNF—Dementia—Diabetes—Alzheimer's disease—Elderly individuals.

Received October 17, 2013; Accepted February 3, 2014 Decision Editor: Rafael de Cabo, PhD

L

IFE expectancy has increased all over the world in the last century (1) leading to an increase in the preva-lence of several chronic diseases. Among these, dementia is one of the most common because its overall incidence is about 2%–4% per year in those older than 75 years, rais-ing to 18% in those older than 90 years (2,3). Research is focused on understanding the mechanisms underlying the pathogenesis of dementia in order to counteract the possi-ble transition from mild cognitive impairment to dementia (4). A particular interest has been addressed to modifiable conditions associated with the pathogenesis of dementia. In humans, both type 1 and type 2 diabetes (T2DM) were associated with impaired cognitive functions including learning memory, attention, and processing speed (5,6). Population-based studies have shown that hyperglycemia and T2DM are associated with an increase in the incidence of cognitive impairment/dementia (7,8), whereas others suggest that both hyperinsulinemia and T2DM give a two-fold increase in risk of Alzheimer’s disease (AD) (9,10). A longer duration of T2DM has been associated with a major risk of cognitive decline (11); moreover, a recent cohort study showed that higher glucose levels may be a

risk factor for dementia, even among nondiabetic individu-als (12). These findings are in line with the observation that T2DM is associated with structural brain abnormali-ties, the hippocampus being the most involved area (13). Several mechanisms have been proposed to explain the association between diabetes and dementia. It was first hypothesized that hypoglycemic state might be responsi-ble of dementia (14). Also, cerebrovascular diseases and related risk factors among diabetic patients might be rel-evant, given the role they play both in vascular dementia (VaD) and late onset Alzheimer’s disease (LOAD) (15–17). Indeed, the combination of T2DM and hypertension gives a higher risk for dementia compared with the sum of the single risk factors (18,19). However, several nonvascular mechanisms have been also considered (20), including for-mation of advanced glycosylation end products (21), brain inflammation (22), alterations of hypothalamus–pituitary adrenal axis with cortisol increase (23), polyol pathway or protein kinase C activation (24), glucose shunting to the exosamine pathway, oxidative stress (25), and disturbed neuronal insulin signaling promoting cerebral amyloidosis (26,27). Interestingly, a high density of insulin receptors March

at Universita degli Studi di Ferrara- Biblioteca S. Maria delle Grazie on March 2, 2016

http://biomedgerontology.oxfordjournals.org/

(2)

was found in brains from LOAD patients, as possible com-pensatory mechanism for reduced insulin functionality) (28); moreover, hyperinsulinemia was demonstrated to reduce degradation and clearance of cerebral Aβ (29).

Brain-derived neurotropic factor (BDNF) is a growth fac-tor member of the neurotropin family; its mature isoform binds specifically to the tropomiosine receptor kinase B (Trk B), a tyrosine kinase receptor, whereas the precursor pro-BDNF binds the pan-neurotropin receptor p75NTR, both mediating different neurotropic signaling (30,31). During development and following insults, BDNF plays a critical role in cell differentiation, migration, neuronal survival, dendritic arborization, synaptogenesis, and synaptic plastic-ity (20). BDNF is also important for learning and memory processes by inducing long-term potentiation in hippocam-pus with structural changes in synapses (32,33). Decreased BDNF concentrations have been found in brains from mild cognitive impairment or LOAD patients. A positive correla-tion between brain BDNF concentracorrela-tion and cognitive per-formance was described (34,35), whereas decreased BDNF production has been proposed as one possible pathogenetic factor for LOAD and major depression (36). Two recent studies investigated serum BDNF levels in patients with dif-ferent neurodegenerative diseases. Whoolley and colleagues (37) found no differences between LOAD, frontotemporal dementia, mild cognitive impairment, and controls, whereas Ventriglia and colleagues (38) reported lower BDNF values in patients with LOAD, VaD, frontotemporal dementia, and Lewy body dementia. Interestingly, plasma BDNF levels are decreased in T2DM and have been inversely correlated with plasma glucose and insulin resistance assessed by homeo-static model assessment. Moreover, plasma BDNF output from human brain is abrogated by hyperglycemia, but is not regulated by hyperinsulinemia (39). In diabetic mice models, BDNF infusion reduces food intake, lowers blood glucose levels, and reduces insulin resistance enhancing insulin peripheral action (40). Moreover, BDNF seems to protect pancreas islets (41), increases insulin, and decreases glucagon pancreatic content (42). Zhen and colleagues (43) found both lower serum BDNF concentrations and cogni-tive functions in diabetic patients versus controls; further-more, a positive relationship between serum BDNF and delayed memory emerged in diabetic patients, suggesting a role of BDNF in cognitive deficit associated with T2DM.

Although T2DM and dementia have been consistently associated in the elderly (7–10), it is not clear whether their correlation with plasma BDNF might be independ-ent or, alternatively, T2DM might mediate the relationship between low BDNF levels and dementia. In order to inves-tigate the possible interplay between BDNF, dementia, and T2DM, we evaluated plasma BDNF levels, according to the presence/absence of T2DM, in a sample of elderly individ-uals including cognitively normal participants, individindivid-uals with cerebrovascular disease but not dementia, and patients affected by LOAD or VaD.

Patients and Methods Participants

During the period 2008–2010, 164 consecutive partici-pants (62.1% women; mean age: 75 ± 10 years) referring to the Day Service for the study of cognitive decline (Institute of Internal Medicine, Gerontology, and Clinical Nutrition; Geriatrics Unit, S. Anna University-Hospital, Ferrara, Italy) were enrolled. Personal data and medical history were col-lected by using a structured interview to patients and car-egivers. All participants (and/or their caregiver if demented) were informed about the research project during the first visit and gave their written consent in order to participate to the study. The study was approved by the local ethic com-mittee (S. Anna University Hospital, Ferrara, Italy) and was conducted in accordance with the Helsinki Declaration as revised in 1989. The participants were divided into four groups based on cognitive status:

1. Forty-four patients with LOAD (mean age 78 ± 8 years; 33 women) by the NINCDS–ADRDA criteria (44). Only patients with “probable” LOAD were included; patients with “possible” LOAD or with LOAD associated with significant cerebrovascular disease on CT scan were excluded in order to increase specificity. The Global Deterioration scale ranged from stage 3 to stage 5. 2. Fifty patients with VaD (mean age: 79 ± 7 years; 24

women) by the NINDS-AIREN criteria (45). Only patients with “probable” VaD were enrolled. The Global Deterioration scale ranged from stage 4 to stage 6. 3. Twenty-three patients with cerebrovascular disease

doc-umented by CT scan and previously affected by transient ischemic attack and/or ischemic stroke, but without evi-dence of dementia (cerebrovascular disease not demen-tia [CDND]; mean age 72 ± 11 years; 11 women). 4. Forty-seven normal individuals without evidence of

cog-nitive impairment (controls: C; mean age 69 ± 10 years, 34 women). All these participants were free-living, healthy (no important comorbidity was found), and independent on basic activities of daily living (BADLs; median Barthel Index score: 98/100). The median Mini-Mental State Examination (MMSE) score was 29/30. Exclusion criteria were participants affected by other types of degenerative dementias (not LOAD), secondary nonvascu-lar dementias, severe liver or kidney disease, severe conges-tive heart failure (New York Heart Association class III–IV), severe chronic obstructive pulmonary disease, cancer, and evidence of acute illnesses at the time of clinical observa-tion. Clinical chemistry analyses were performed in order to exclude secondary cognitive impairment. No cerebrospinal fluid biomarkers were available for all the patients (LOAD and VaD) enrolled into this study. All patients underwent a general and neurological examination. The diagnosis of dementia was made by trained geriatricians. For neuropsy-chological assessment, all patients were given a battery of

(3)

tests evaluating: verbal memory (Rey’s 15-word test), work-ing memory (digit span forward–backward), prose memory (Babcock test), space/time orientation (items from MMSE), attention (Tolouse–Pieron test), constructional and visuospa-tial functions (clock-drawing test), abstract reasoning (Raven progressive matrices, similarities test), language and compre-hension (Token test), verbal fluency (letters and categories), executive functions (Trial-making test A and B), and routine clinical tests for the evaluation of agnosia, apraxia, and afa-sia. Depressive symptoms were evaluated by the Geriatric Depression scale (GDS). Functional dependence was evalu-ated by Barthel’s Index for BADL, and Lawton–Brody modi-fied index for instrumental activities of daily living.

Participants were further classified according to presence/ absence of T2DM. All participants with known history of diabetes or current hypoglycaemic therapy were defined as diabetic patients. Criteria of the American Diabetes Association were used for new diagnosis: fasting plasma glucose ≥ 126 mg/dL (7.0 mmol/L) or 2-hour plasma glu-cose ≥ 200 mg/dL (11.1 mmol/L) during oral gluglu-cose toler-ance test (46). The glycated hemoglobin (HbA1C) ≥ 6.5% was not used as diabetes diagnostic criteria because it was not available in all participants. Diabetic participants were 37 (mean age 76 ± 9 years), 122 had no diabetes (mean age 74 ± 10), and 5 individuals were not classified with certainty.

The criteria for the diagnosis of hypertension were (a) history of known hypertension or antihypertensive therapy at visit time and (b) blood pressure > 140/90 mmHg in three or more measurements. Eighty patients were affected by hypertension. No patients were taking a statin at the time of enrolment into the study.

Brain CT Scan

All patients (LOAD, VaD, CDND) underwent a brain CT. The instrument used was a third generation SIEMENS SOMATON HQ. The slice thickness was 10 mm. Radiograms were evaluated by two trained radiologists not informed about the patient. The CT scan information sup-ported the diagnosis and excluded other brain pathologies.

Sample Processing and Analytical Methods

Blood samples were drawn from a forearm vein in the morning after overnight fasting. All samples were kept chilled in an ice bath until centrifugation at 3000 rpm for 15 min at 4°C. The separated plasma was stored at −80°C until time of assay. All samples were run in duplicate for the same assay. Glucose concentrations were measured with the glucose oxidase technique using an auto ana-lyzer (ILAB600, Instrumentation Laboratories S.p.A., Milano, Italy). The minimum detectable concentrations was 0.11 mmol/L (2 mg/mL). Intra- and interassay variation coefficients are 3.0% and 3.5%, respectively. Serum total cholesterol and triglycerides levels were assayed by the Trinder method. Serum HDL-C levels were measured after

precipitation of the Apo B-containing lipoproteins by add-ing phosphotungstic acid and magnesium ions to the sample (47). LDL cholesterol (LDL-C) levels were calculated by the Friedewald’s formula (LDL-C = total cholesterol − tri-glycerides/5 − HDL-C) (48).

Immunoassay Systems (Promega) specific for BDNF were performed according to the manufacturer’s instruc-tions. The minimum detectable concentrations is 15.6 pg/mL. Intra- and interassay variation coefficients are 3.0% and 5.0%, respectively.

Statistical Analysis

Continuous variables were expressed as mean (standard deviation) or median (interquartile range), whereas cat-egorical variables were expressed as the number/percentage. Mean values were compared by ANOVA with Fisher’s least significant difference (LSD) post hoc test for multiple com-parison, whereas medians were compared by nonparamet-ric tests (Kruskal–Wallis). Correlations between continuous variables were tested by Pearson’s correlation or Spearman’s test when necessary. Prevalence was compared by the χ2_test.

Multivariate linear regression analysis was used to test the independent association between BDNF serum levels and other variables of interest; diagnosis of dementia or diabetes were entered as dichotomous variables (absent: 0; present: 1). All statistical tests were performed using SPSS 17.0 software (SPSS Inc., Chicago, IL).

Results

In Table 1 are reported the principal characteristics

of the four groups of individuals enrolled into the study. Individuals with dementia were older compared with those without dementia. The prevalence of female gender was higher in LOAD and controls compared with CDND and VaD. A trend toward higher GDS score was observed in demented compared with nondemented individuals (not significant). The prevalence of diabetes was higher in participants with cerebrovascular disease (both VaD and CDND) and in LOAD compared with controls, whereas no differences emerged as regards blood glucose levels. The prevalence of hypertension was also significantly higher in participants with cerebrovascular disease (VaD and CDND) compared with the other two groups.

Plasma BDNF levels were significantly lower in diabetic patients (276.4 ± 129.2 pg/mL) compared with nondiabetic patients (341.7 ± 157.6; ANOVA, LSD post hoc test: p: .02), as well as in demented (LOAD + VaD: 290.5 ± 132.7) com-pared with nondemented participants (CDND + C: 381.7 ± 182.6 pg/mL; ANOVA, LSD post hoc test p: .001).

As shown in Figure 1, the mean levels of BDNF progres-sively decreased from C (383.9 ± 204.6 pg/mL), to CDND (377.1 ± 130.2 pg/mL), to VaD (313.3 ± 114.8 pg/mL), to LOAD (264.7 ± 147.7 pg/mL). LOAD displayed signifi-cantly lower BDNF compared with C and CDND (ANOVA,

(4)

LSD post hoc test: p: .002 and p: .03, respectively). A similar pattern was observed when the sample was divided accord-ing to the presence/absence of T2DM. Among nondiabetic patients, LOAD had lower BDNF compared with C and CDND (ANOVA, LSD post hoc test: p: .05 and p: .03) (C: 380.1 ± 188.6 pg/mL; CDND: 421.7 ± 120.3 pg/mL; VaD: 311.5 ± 123.5 pg/mL; LOAD: 290.5 ± 138.5 pg/mL). Among

diabetic patients, LOAD had lower BDNF compared with controls and VaD (ANOVA, LSD post hoc test p: .04 and p: .032, respectively; C: 364.2 ± 105.4; CDND: 293.3 ± 109.7; VaD: 316.7 ± 98.9; LOAD: 176.8 ± 150.8 pg/mL).

Finally, the sample was divided into four groups accord-ing to the absence/presence of diabetes and dementia

(Figure 2): group 1: no diabetes nor dementia (n: 60); group

Figure 1. Brain-derived neurotrophic factor plasma levels in healthy controls, participants with cerebrovascular disease not dementia, vascular dementia, and late onset Alzheimer’s disease (ANOVA, least significant difference post hoc test).

Table 1. Principal Characteristics of Elderly Patients With Cerebrovascular Disease Not Dementia (CDND), Vascular Dementia (VaD), Late Onset Alzheimer’s Disease (LOAD), and Controls (means ± standard deviation or median–interquartile range)

Dementia (n = 94) No Dementia (n = 70)

All (n = 164)

VaD (n = 50) LOAD (n = 44) CDND (n = 23) Controls (n = 47)

Age (y) 79 ± 7 79 ± 8 72 ± 11**,§ _{67 ± 10***},§§ _{75 ± 10} Female gender (n/%)^ _24/48.0 _33/75.0 _11/47.8 _34/72.3 _102/62.2 BADLs (/100) 6/100 (1–16) 17/100 (11–19) 75/100***,§§_(48–100) _100/100***,§§_(96–100) _{18/100 (6–96)} MMSE score (/30) 17.5 (11.5–20) 17 (12.5–19) 30***,§§_(27–30) _30***,§§_(30–30) _{23 (20–28)} GDS score (/15)+ _{7 (3.5–13.5)} _{9 (4–11.5)} _{5 (4–9)} _{5 (1.5–6.5)} _{6 (4–11)} Diabetes (n/%)^^^ _17/34.0 _10/22.7 _8/34.8 _2/4.3 _37/22.6 Glucose (mg/dL) 102 (97–109) 94 (89–120) 98 (88–118) 100 (93–120) 98 (90–119) Hypertension (n/%)^^ _33/66.0 _16/36.4 _13/56.5 _18/38.3 _80/48.8 SBP (mmHg) 141 ± 18 136 ± 21 143 ± 17 130 ± 16# _{137 ± 18} DBP (mmHg) 81 ± 7 79 ± 10 81 ± 7 81 ± 8 81 ± 8 Total cholesterol (mg/dL) 199 ± 53 221 ± 53 209 ± 55 221 ± 32 213 ± 49 LDL cholesterol (mg/dL) 117 ± 36 138 ± 42 134 ± 45 145 ± 31* 134 ± 38 HDL cholesterol (mg/dL) 44 ± 15 53 ± 13 43 ± 12 54 ± 13* 50 ± 14 Triglycerides (mg/dL) 119 (117–172) 103 (88–120) 124 (109–220) 106 (76–108) 108 (99–172) Hematocrit (%) 40 ± 8 39 ± 5 38 ± 4 40 ± 4 39 ± 7

White blood cells (/mm3₎ _{6817 ± 2188} _{6689 ± 2234} _{6633 ± 1438} _{6048 ± 1627} _{6635 ± 2023}

Notes: BADLs = basis aactivities of daily living; MMSE = Mini-Mental State Examination; GDS = Geriatric Depression scale; SBP = systolic blood pressure; DBP = diastolic blood pressure.

*vs VaD p < .05; **vs VaD p < .01; ***vs VaD p < .001.

§_{vs LOAD p < .05;}§§_{vs LOAD p < .001.}

#_{vs CDND p < .05.}

^_{Chi square test p < .01;}^^_{Chi square test p = .05;}^^^_{Chi square test p < .005.}

+_{.06 Kruskal–Wallis.} at Universita degli Studi di Ferrara- Biblioteca S. Maria delle Grazie on March 2, 2016

(5)

2: only dementia (n: 94); group 3: only diabetes (n: 10); and group 4: both diabetes and dementia (n: 27). Compared with group 1 (391 ± 171 pg/mL), plasma BDNF was sig-nificantly lower in group 4 (264.9 ± 136 pg/mL) and group 2 (301 ± 130.7 pg/mL; ANOVA, LSD post hoc test p: .001 and p: .001, respectively), whereas the difference was not significant for group 3 (307 ± 107 pg/mL), probably due to the small number of individuals with diabetes only. No sig-nificant differences in BDNF levels emerged between men and women, nor between nonhypertensive and hypertensive individuals (data not shown).

We also evaluated the principal correlate of plasma BDNF levels. At univariate analysis, BDNF was negatively correlated with age and systolic blood pressure, and posi-tively correlated with MMSE, GDS, total cholesterol, and LDL-C levels (Table 2). By means of multivariate linear regression analysis, we found that dementia (β coefficient: −241.8; p: .006) and diabetes (β coefficient:-100.1: p:0.05) significantly predicted BDNF levels, independent of pos-sible confounders including age, MMSE, GDS, total cho-lesterol, and systolic blood pressure (R2_{for adjusted model:}

.41; p: .001; Table 3).

Discussion

We found that in a sample of elderly individuals, both dementia and diabetes were independently associated with lower BDNF plasma levels. Moreover, the effect of

dementia and diabetes on BDNF was “addictive” because BDNF levels were significantly lower in patients affected by dementia and diabetes, compared with both diabetic controls and nondiabetic, demented individuals.

BDNF and Diabetes

Besides our study, a negative correlation between BDNF levels and diabetes has been consistently reported, but the direction of the association is unclear. The cerebral output of BDNF is inhibited by hyperglycemia, and this might explain the association between low BDNF levels and insulin resist-ance (25). On the other hand, BDNF decrease might con-tribute to the pathogenesis of diabetes because it has been shown, in animal and in vitro studies, that BDNF might have

Figure 2. Brain-derived neurotrophic factor plasma levels in 164 older individuals according to absence/presence of diabetes mellitus and dementia (ANOVA, least significant difference post hoc test).

Table 2. Correlations Between Brain-Derived Neurotrophic Factor Plasma Levels and Other Variables in 164 Older Individuals

Correlation Coefficient p Age (y) −0.22 .004 MMSE score (/30) 0.33 .001 GDS score (/15) 0.31 .008 Glucose (mg/dL) −0.07 .38 SBP (mmHg) −0.19 .02 DBP (mmHg) −0.09 .25 Total cholesterol (mg/dL) 0.23 .004 HDL cholesterol (mg/dL) 0.04 .61 LDL cholesterol (mg/dL) 0.19 .04 Triglycerides (mg/dL) 0.09 .22

(6)

antidiabetic effects (41,42). A recent study suggests that, in animal models, the brain intraventricular administration of BDNF attenuates diabetic hyperglycemia via an insulin-independent mechanism involving inhibition of glucagon secretion and decrease in hepatic glucose production (49). In this light, a sort of “positive feedback relationship” might be hypothesized between BDNF and diabetes. As regards the prevalence of T2DM in our sample (22.6%), it is in line with the results of a Greek study estimating a prevalence of diabe-tes among elderly individuals about 22% (50), but it is much higher than that calculated by a British study (7%) among people older than 75 years (51). The finding of a higher prev-alence of T2DM in demented patients (28%) compared with controls (4.3%) indirectly support the concept that diabetes might increase the risk of developing dementia through mul-tiple mechanisms (7,8,11).

BDNF and Dementia

On the whole, demented patients had lower BDNF levels compared with nondemented patients; moreover, the reduc-tion of BDNF levels showed a sort of “progressive pattern” decreasing from controls, to CDND, to VaD, to LOAD patients. A complicated relationship has been reported in Literature between BDNF levels and dementia. O’Bryant and colleagues found that among ApoE4 negative AD patients, increased BDNF levels were associated with poorer performances at visual–verbal memory tests, and hypoth-esized an upregulation of BDNF as possible compensa-tory mechanism (52). Leyhe and colleagues found reduced BDNF levels in AD patients compared with controls and demonstrated a significant increase in BDNF concentration after treatment with Donepezil (53). Interestingly, Laske and colleagues (54) found that BDNF was increased in early AD compared with both late AD patients and controls, leading to the hypothesis of an initial increase of plasma BDNF (pos-sible compensatory mechanism), followed by a later BDNF decrease with lack of trophic support. This hypothesis is

in line with in vitro studies in which BDNF protects neu-ronal cultures against cytotoxic effects of beta-amyloid (55), whereas sublethal doses of beta-amyloid downregulate BDNF expression in cortical neurons (56) but upregulate BDNF production in astrocytes (57). In this light, our results are in good agreement with literature results; indeed, by comparing MMSE scores of LOAD patients from our and Laske’s work (39), it is evident that our patients might be considered in an “advanced” stage of dementia. The lack of significant difference in serum BDNF between LOAD and VaD in our sample is in line with work of Ventriglia and colleagues (38) and supports the concept that low BDNF levels might be a nonspecific marker of neurodegeneration. BDNF is interested early in the development of dementia, as supported by the results from the Framingham study; each standard deviation increment in serum BDNF was associated with a 33% lower risk of developing dementia and LOAD on 10-year follow-up (58). However, the role of BDNF in the pathogenesis of LOAD is not clear yet. In one hand, low BDNF might be a consequence of amyloid deposition because it was demonstrated in AD mice that Aβ oligomer deposition compromises BDNF retro-trafficking by reducing the endosomial vesicles velocity (59). An early involvement of BDNF/pro-BDNF has been also demonstrated, caused by Aβ action on cyclic-AMP response element binding protein phosphorilation, with consequent reduction in BDNF gene expression (60). Allen and colleagues (61) postulated that Aβ deposition should be the “primummovens,” whereas consequent BDNF reduction should initiate a cascade of events exacerbating the pathology and leading to dementia. In the other hand, it was reported that BDNF is an induc-tor of SORLA (sorting protein-related recepinduc-tor containing LDLR class A repeats) transcription (62), a molecule regu-lating amyloid precursor protein intracellular trafficking and processing into Aβ, which reduces amyloid plaque forma-tion when overexpressed (63,64).This observation suggests that BDNF reduction might be the first step of the process, inducing the accumulation of Aβ.

BDNF, Hypertension, and Cholesterol Levels

In this study, a negative correlation between systolic blood pressure and BDNF levels and a positive correlation between LDL-C and BDNF levels were found at univariate analysis; however, at multivariate analysis, neither systolic blood pressure nor LDL-C predicted BDNF levels, suggest-ing that other factors included into the model (eg, diabetes, age) might mediate these correlations.

The relationship between serum BDNF, blood pressure, and other metabolic parameters was evaluated by Golden and colleagues (65); a positive correlation of BDNF levels with diastolic blood pressure was found in men and women, whereas a positive correlation with total cholesterol/LDL-C emerged only in women. Because plasma BDNF decreases with age (66), whereas the prevalence of hypertension and Table 3. Multivariate Linear Regression Analysis for Brain-Derived

Neurotrophic Factor Plasma Levels in 164 Older Individuals (R2_for

the adjusted model: 0.41; p: .001)

Model

Unstandardized Coefficients

p

Beta Standard Error

Constant 549.2 261.1 .04 Age −3.06 2.94 .30 Dementia −241.8 81.8 .007 Diabetes −100.1 50.7 .03 MMSE score 9.3 7.7 .23 GDS score 12.4 3.9 .003 Total cholesterol (mg/dL) 0.7 0.3 .07 SBP (mmHg) −0.6 1.4 .65

Notes: Dementia and diabetes were entered as categorical variables (absent: 0; present: 1). Values in bold underline the variables whose p value reached statistical significance.

(7)

metabolic syndrome increases, these authors concluded that BDNF might contribute to lipids and blood pressure regu-lation and that their finding might represent a compensa-tory response to disrupted lipid metabolism and increase in blood pressure. A more recent study (67) analyzed BDNF levels in heart and aorta of hypertensive versus normoten-sive rats, finding a reduced local expression in the former; moreover, administration of exogenous BDNF induced aor-tic dilation, confirming the role of BDNF in the regulation of endothelial function.

BDNF and Depressive Symptoms

A significant relationship between GDS score and BDNF plasma levels emerged from our study. In particular, GDS positively correlated with BDNF levels, suggesting that BDNF levels were higher in patients complaining more depressive symptoms. This positive correlation emerged in the whole sample but was confirmed both in demented and nondemented individuals (data not shown). Some stud-ies have reported reduced BDNF levels in patients with major depression (68,69), and it has been suggested that the reduction of BDNF might contribute to AD and major depression pathogenesis (24). Only a few studies have investigated the relationship between depressive symptoms and BDNF in AD patients. Consistent with our results, Hall and colleagues found significantly higher BDNF lev-els in depressed compared with nondepressed AD patients (70); these authors hypothesized that the increase in BDNF might reflect a chronic inflammatory process related to the pathophysiology of depression (45). Other studies showed different results: Laske and colleagues found a negative correlation between BDNF plasma concentrations and depressive symptoms measured by GDS (71), whereas Lee and colleagues found no differences between depressed and nondepressed AD patients (72). However, in the first study, the number of AD patients assessed was small, whereas in the second, the GDS cutoff for severe depression was high (score ≥ 20/30) so that many of the patients considered “not severely depressed” (and thus excluded) would have met the criteria for depression in our study.

Finally, we have to acknowledge some important limita-tions of the study. First, the size of the whole sample, and consequently of the four subgroups, was not really large, thus limiting the statistical power of the study. Second, the cross-sectional design does not allow to advance any cause-effect relationships. Consequently, we do not know whether reduced BDNF levels might be the “primummovens,” inducing both diabetes and dementia or its decrease might result from the presence of diabetes and/or dementia. However, although further longitudinal studies on this topic are needed, our results clearly confirm the existence of an independent relationship between plasma BDNF, diabetes, and dementia among elderly people. Third, we were not aware about the possible antidepressant treatment of our

patients, and antidepressant therapies are known to increase serum BDNF in the general population (73,74) and in VaD (75). At last, another limitation of the study regards the physical activity level. In animals, physical activity increases BDNF concentration in the hippocampus (76); moreover, a positive correlation between physical activity and serum BDNF was reported in AD (77). The increase of serum BDNF during physical exercise seems to be due to enhanced brain release because the brain contributes to 70%–80% of circulating BDNF (78). The degree of physi-cal activity was not estimated in our study; nevertheless, indirect information came from the evaluation of BADLs, which express the functional autonomy of the individual. It is evident that patients with low BADLs score perform very limited physical activity. There was a significant difference in BADLs between demented and not demented individuals (see Table 1), the demented participants having the lower scores. Thus, our results might depend, at least in part, from the effect of different degree of physical activity. However, dementia definition requires a loss of function from the previous normal state; of consequence, very often patients with dementia display lower BADLs score compared with healthy elderly participants.

Conclusion

In conclusion, we found that among elderly individu-als, both diagnoses of dementia and diabetes mellitus were associated with lower levels of plasma BDNF and correlated independently with BDNF plasma levels. Demented patients also affected by diabetes had the low-est BDNF levels in the sample, sugglow-esting the existence of a “synergistic” effect of dementia and diabetes on BDNF levels.

References

1. Kinsella K, Wan H. U.S. Census Bureau: International Population Reports, P95/09-1, an Aging World: 2008. Washington, DC: U.S. Government Printing Office; 2009.

2. Corrada MM, Brookmeyer R, Paganini-Hill A, Berlau D, Kawas CH. Dementia incidence continues to increase with age in the oldest old: the 90+ study. Ann Neurol. 2010;67:114–121. doi:10.1002/ana.21915 3. Rocca WA. Dementia, Parkinson’s disease, and stroke in Europe:

A commentary. Neurology. 2000;54(suppl 5):S38–S40.

4. Querfurth HW, LaFerla FM. Alzheimer’s disease. N Engl J Med. 2010;362:329–344. doi:10.1056/NEJMra0909142

5. Awad N, Gagnon M, Messier C. The relationship between impaired glucose tolerance, type 2 diabetes, and cognitive function. J Clin Exp Neuropsychol. 2004;26:1044–1080. doi:10.1080/13803390490514875 6. Kodl CT, Seaquist ER. Cognitive dysfunction and diabetes mellitus.

Endocr Rev. 2008;29:494–511. doi:10.1210/er.2007-0034

7. Allen KV, Frier BM, Strachan MW. The relationship between type 2 diabetes and cognitive dysfunction: longitudinal studies and their methodological limitations. Eur J Pharmacol. 2004;490:169–175. doi:10.1016/j.ejphar.2004.02.054

8. MacKnight C, Rockwood K, Awalt E, McDowell I. Diabetes melli-tus and the risk of dementia, Alzheimer’s disease and vascular cogni-tive impairment in the Canadian Study of Health and Aging. Dement Geriatr Cogn Disord. 2002;14:77–83. doi:10.1159/000064928

(8)

9. Ott A, Stolk RP, van Harskamp F, Pols HA, Hofman A, Breteler MM. Diabetes mellitus and the risk of dementia: The Rotterdam Study. Neurology. 1999;53:1937–1942. doi:10.1212/WNL.53.9.1937 10. Luchsinger JA, Tang MX, Shea S, Mayeux R. Hyperinsulinemia

and risk of Alzheimer disease. Neurology. 2004;63:1187–1192. doi:10.1212/01.WNL.0000140292.04932.87

11. Gregg EW, Yaffe K, Cauley JA, et al. Is diabetes associated with cognitive impairment and cognitive decline among older women? Study of Osteoporotic Fractures Research Group. Arch Intern Med. 2000;160:174–180. doi:10.1001/archinte.160.2.174

12. Crane PK, Walker R, Hubbard RA, et al. Glucose levels and risk of dementia. N Engl J Med. 2013;369:540–548. doi:10.1056/ NEJMoa1215740

13. Bruehl H, Wolf OT, Sweat V, Tirsi A, Richardson S, Convit A. Modifiers of cognitive function and brain structure in middle-aged and elderly individuals with type 2 diabetes mellitus. Brain Res. 2009;1280:186–194. doi:10.1016/j.brainres.2009.05.032

14. Perros P, Deary IJ. Long-term effects of hypoglycaemia on cogni-tive function and the brain in diabetes. In: Frier BM, Fisher BM, eds. Hypoglycaemia in Clinical Diabetes. Chichester, UK: Wiley; 1999:187–210.

15. De la Torre JC. Alzheimer disease as a vascular disorder: nosologi-cal evidence. Stroke. 2002;33:1152–1162. doi:10.1161/01. STR.0000014421.15948.67

16. Newman AB, Fitzpatrick AL, Lopez O, et al. Dementia and Alzheimer’s disease incidence in relationship to cardiovascular dis-ease in the Cardiovascular Health Study cohort. J Am Geriatr Soc. 2005;53:1101–1107. doi:10.1111/j.1532-5415.2005.53360.x 17. van Oijen M, de Jong FJ, Witteman JC, Hofman A, Koudstaal PJ,

Breteler MM. Atherosclerosis and risk for dementia. Ann Neurol. 2007;61:403–410. doi:10.1002/ana.21073

18. Xu WL, Qiu CX, Wahlin A, Winblad B, Fratiglioni L. Diabetes melli-tus and risk of dementia in the Kungsholmen project: a 6-year follow-up study. Neurology. 2004;63:1181–1186.

19. Whitmer RA, Sidney S, Selby J, Johnston SC, Yaffe K. Midlife car-diovascular risk factors and risk of dementia in late life. Neurology. 2005;64:277–281.

20. Cole AR, Astell A, Green C, Sutherland C. Molecular connex-ions between dementia and diabetes. Neurosci Biobehav Rev. 2007;31:1046–1063. doi:10.1016/j.neubiorev.2007.04.004

21. Sato T, Shimogaito N, Wu X, Kikuchi S, Yamagishi S, Takeuchi M. Toxic advanced glycation end products (TAGE) theory in Alzheimer’s disease. Am J Alzheimers Dis Other Demen. 2006;21:197–208. doi:10.1177/1533317506289277

22. Yaffe K, Blackwell T, Whitmer RA, Krueger K, Barrett Connor E. Glycosylated hemoglobin level and development of mild cogni-tive impairment or dementia in older women. J Nutr Health Aging. 2006;10:293–295.

23. Tojo C, Takao T, Nishioka T, Numata Y, Suemaru S, Hashimoto K. Hypothalamic-pituitary-adrenal axis in WBN/Kob rats with non-insulin dependent diabetes mellitus. Endocr J. 1996;43:233–239. doi:10.1507/endocrj.43.233

24. Ramakrishnan R, Sheeladevi R, Suthanthirarajan N. PKC-alpha medi-ated alterations of indoleamine contents in diabetic rat brain. Brain Res Bull. 2004;64:189–194. doi:10.1016/j.brainresbull.2004.07.002 25. Biessels GJ, Kappelle AC, Bravenboer B, Erkelens DW, Gispen

WH. Cerebral function in diabetes mellitus. Diabetologia. 1994;37:643–650.

26. Carro E, Torres-Aleman I. The role of insulin and insulin-like growth factor I in the molecular and cellular mechanisms underlying the pathology of Alzheimer’s disease. Eur J Pharmacol. 2004;490:127– 133. doi:10.1016/j.ejphar.2004.02.050

27. Kodl CT, Franc DT, Rao JP, et al. Diffusion tensor imaging identifies deficits in white matter microstructure in subjects with type 1 dia-betes that correlate with reduced neurocognitive function. Diadia-betes. 2008;57:3083–3089. doi:10.2337/db08-0724

28. Frölich L, Blum-Degen D, Bernstein HG, et al. Brain insulin and insulin receptors in aging and sporadic Alzheimer’s disease. J Neural Transm. 1998;105:423–438.

29. Craft S, Watson GS. Insulin and neurodegenerative disease: shared and specific mechanisms. Lancet Neurol. 2004;3:169–178. doi:10.1016/ S1474-4422(04)00681-7

30. Cohen-Cory S, Kidane AH, Shirkey NJ, Marshak S. Brain-derived neurotrophic factor and the development of structural neuronal con-nectivity. Dev Neurobiol. 2010;70:271–288. doi:10.1002/dneu.20774 31. Zoladz JA, Pilc A. The effect of physical activity on the brain

derived neurotrophic factor: from animal to human studies. J Physiol Pharmacol. 2010;61:533–541.

32. Tyler WJ, Alonso M, Bramham CR, Pozzo-Miller LD. From acqui-sition to consolidation: on the role of brain-derived neurotrophic factor signaling in hippocampal-dependent learning. Learn Mem. 2002;9:224–237. doi:10.1101/lm.51202

33. Allen SJ, Watson JJ, Dawbarn D. The neurotrophins and their role in Alzheimer’s disease. Curr Neuropharmacol. 2011;9:559–573. doi:10.2174/157015911798376190

34. Peng S, Wuu J, Mufson EJ, Fahnestock M. Precursor form of brain-derived neurotrophic factor and mature brain-derived neurotrophic factor are decreased in the pre-clinical stages of Alzheimer’s disease. J Neurochem. 2005;93:1412–1421. doi:10.1111/j.1471-4159.2005.03135.x

35. Connor B, Young D, Yan Q, Faull RL, Synek B, Dragunow M. Brain-derived neurotrophic factor is reduced in Alzheimer’s disease. Brain Res Mol Brain Res. 1997;49:71–81.

36. Tsai SJ. Brain-derived neurotrophic factor: a bridge between major depression and Alzheimer’s disease? Med Hypotheses. 2003;61:110– 113. doi:10.1016/S0306-9877(03)00141-5

37. Woolley JD, Strobl EV, Shelly WB, et al. BDNF serum concentra-tions show no relaconcentra-tionship with diagnostic group or medication status in neurodegenerative disease. Curr Alzheimer Res. 2012;9:815–821. doi:10.2174/156720512802455395

38. Ventriglia M, Zanardini R, Bonomini C, et al. Serum brain-derived neurotrophic factor levels in different neurological diseases. Biomed Res Int. 2013;2013:901082. doi:10.1155/2013/901082

39. Krabbe KS, Nielsen AR, Krogh-Madsen R, et al. Brain-derived neurotrophic factor (BDNF) and type 2 diabetes. Diabetologia. 2007;50:431–438. doi:10.1007/s00125-006-0537-4

40. Ono M, Ichihara J, Nonomura T, et al. Brain-derived neurotrophic factor reduces blood glucose level in obese diabetic mice but not in normal mice. Biochem Biophys Res Commun. 1997;238:633–637. doi:10.1006/bbrc.1997.7220

41. Yamanaka M, Itakura Y, Inoue T, et al. Protective effect of brain-derived neurotrophic factor on pancreatic islets in obese dia-betic mice. Metabolism. 2006;55:1286–1292. doi:10.1016/j. metabol.2006.04.017

42. Yamanaka M, Itakura Y, Tsuchida A, Nakagawa T, Noguchi H, Taiji M. Comparison of the antidiabetic effects of brain-derived neurotrophic factor and thiazolidinediones in obese diabetic mice. Diabetes Obes Metab. 2007;9:879–888. doi:10.1111/j.1463-1326.2006.00675.x 43. Zhen YF, Zhang J, Liu XY, et al. Low BDNF is associated with

cog-nitive deficits in patients with type 2 diabetes. Psychopharmacology (Berl). 2013;227:93–100. doi:10.1007/s00213-012-2942-3

44. McKhann G, Drachman D, Folstein M, Katzman R, Price D, Stadlan EM. Clinical diagnosis of Alzheimer’s disease: report of the NINCDS-ADRDA Work Group under the auspices of Department of Health and Human Services Task Force on Alzheimer’s Disease. Neurology. 1984;34:939–944.

45. Román GC, Tatemichi TK, Erkinjuntti T, et al. Vascular dementia: diagnostic criteria for research studies. Report of the NINDS-AIREN International Workshop. Neurology. 1993;43:250–260.

46. American Diabetes Association. Diagnosis and classification of diabe-tes mellitus. Diabediabe-tes Care. 2013;36(Suppl 1):S67–S74. doi:10.2337/ dc13-S067.

(9)

47. Burstein M, Scholnick HR, Morfin R. Rapid method for the isolation of lipoproteins from human serum by precipitation with polyanions. J Lipid Res. 1970;11:583–595.

48. Friedewald WT, Levy RI, Fredrickson DS. Estimation of the concen-tration of low-density lipoprotein cholesterol in plasma, without use of the preparative ultracentrifuge. Clin Chem. 1972;18:499–502. 49. Meek TH, Wisse BE, Thaler JP, et al. BDNF action in the brain

attenuates diabetic hyperglycemia via insulin-independent inhibi-tion of hepatic glucose producinhibi-tion. Diabetes. 2013;62:1512–1518. doi:10.2337/db12-0837

50. Tyrovolas S, Zeimbekis A, Bountziouka V, et al. Factors associated with the prevalence of diabetes mellitus among elderly men and women living in Mediterranean Islands: the MEDIS Study. Rev Diabet Stud. 2009;6:54–63. doi:10.1900/RDS.2009.6.54

51. Hewitt J, Smeeth L, Bulpitt CJ, Fletcher AE. The prevalence of Type 2 diabetes and its associated health problems in a commu-nity-dwelling elderly population. Diabet Med. 2009;26:370–376. doi:10.1111/j.1464-5491.2009.02687.x

52. O’Bryant SE, Hobson VL, Hall JR, et al.; Texas Alzheimer’s Research Consortium. Serum brain-derived neurotrophic factor levels are spe-cifically associated with memory performance among Alzheimer’s disease cases. Dement Geriatr Cogn Disord. 2011;31:31–36. doi:10.1159/000321980

53. Leyhe T, Stransky E, Eschweiler GW, Buchremer G, Laske C. Increase of BDNFserumconcentration during donepeziltreatment of patients with earlyAlzheimer’sdisease. Eur Arch Psychiatry Clin Neurosci. 2008;258(2):124–128. doi:10.1007/s00406-007-0764-9

54. Laske C, Stransky E, Leyhe T, et al. Stage-dependent BDNF serum concentrations in Alzheimer’s disease. J Neural Transm. 2006;113:1217–1224. doi:10.1007/s00702-005-0397-y

55. Aliaga E, Silhol M, Bonneau N, Maurice T, Arancibia S, Tapia-Arancibia L. Dual response of BDNF to sublethal concentrations of beta-amyloid peptides in cultured cortical neurons. Neurobiol Dis. 2010;37:208–217. doi:10.1016/j.nbd.2009.10.004

56. Poon WW, Blurton-Jones M, Tu CH, et al. β-Amyloid impairs axonal BDNF retrograde trafficking. Neurobiol Aging. 2011;32:821–833. doi:10.1016/j.neurobiolaging.2009.05.012

57. Kimura N, Takahashi M, Tashiro T, Terao K. Amyloid beta up-reg-ulates brain-derived neurotrophic factor production from astrocytes: rescue from amyloid beta-related neuritic degeneration. J Neurosci Res. 2006;84:782–789. doi:10.1002/jnr.20984

58. Weinstein G, Beiser AS, Hoan Choi S, et al. Serum brain-derived neurotrophic factor and the risk for dementia. The Framingham heart study. JAMA Neurol. 2014;71:55–61. doi:10.1001/ jamaneurol.2013.4781

59. Poon WW, Carlos AJ, Aguilar BL, et al. β-Amyloid (Aβ) oligom-ers impair brain-derived neurotrophic factor retrograde trafficking by down-regulating ubiquitin C-terminal hydrolase, UCH-L1. J Biol Chem. 2013;288:16937–16948. doi:10.1074/jbc.M113.463711 60. Arancio O, Chao MV. Neurotrophins, synaptic plasticity and

dementia. Curr Opin Neurobiol. 2007;17:325–330. doi:10.1016/j. conb.2007.03.013

61. Allen SJ, Watson JJ, Dawbarn D. The neurotrophins and their role in Alzheimer’s disease. Curr Neuropharmacol. 2011;9:559–573. doi:10.2174/157015911798376190

62. Rohe M, Synowitz M, Glass R, Paul SM, Nykjaer A, Willnow TE. Brain-derived neurotrophic factor reduces amyloidogenic pro-cessing through control of SORLA gene expression. J Neurosci. 2009;29:15472–15478. doi:10.1523/JNEUROSCI.3960-09.2009

63. Andersen OM, Reiche J, Schmidt V, et al. Neuronal sorting protein-related receptor sorLA/LR11 regulates processing of the amyloid pre-cursor protein. Proc Natl Acad Sci U S A. 2005;102:13461–13466. doi:10.1073/pnas.0503689102

64. Offe K, Dodson SE, Shoemaker JT, et al. The lipoprotein receptor LR11 regulates amyloid beta production and amyloid precursor pro-tein traffic in endosomal compartments. J Neurosci. 2006;26:1596– 1603. doi:10.1523/JNEUROSCI.4946-05.2006

65. Golden E, Emiliano A, Maudsley S, et al. Circulating brain-derived neurotrophic factor and indices of metabolic and cardiovascular health: data from the Baltimore Longitudinal Study of Aging. PLoS One. 2010;5:e10099. doi:10.1371/journal.pone.0010099

66. Lommatzsch M, Zingler D, Schuhbaeck K, et al. The impact of age, weight and gender on BDNF levels in human platelets and plasma. Neurobiol Aging. 2005;26:115–123. doi:10.1016/j. neurobiolaging.2004.03.002

67. Prigent-Tessier A, Quirié A, Maguin-Gaté K, et al. Physical training and hypertension have opposite effects on endothelial brain-derived neurotrophic factor expression. Cardiovasc Res. 2013;100:374–382. doi:10.1093/cvr/cvt219

68. Brunoni AR, Lopes M, Fregni F. A systematic review and meta-analysis of clinical studies on major depression and BDNF lev-els: implications for the role of neuroplasticity in depression. Int J Neuropsychopharmacol. 2008;11:1169–1180. doi:10.1017/ S1461145708009309

69. Karege F, Perret G, Bondolfi G, Schwald M, Bertschy G, Aubry JM. Decreased serum brain-derived neurotrophic factor levels in major depressed patients. Psychiatry Res. 2002;109:143–148. doi:10.1016/ S0165-1781(02)00005-7

70. Hall JR, O’Bryant SE, Johnson L, Barber RC. Depression and brain-derived neurotrophic factor levels in Alzheimer’s disease. Neurosci Med. 2011;2:43–47. doi:10.1159/000321980

71. Laske C, Stransky E, Eschweiler G, et al. BDNF serum concentrations in patientswith Alzheimer’s diseaseare associated with depressive mood states. Neurol Psychiatr Brain Res. 2005;12:1–4.

72. Lee JG, Shin BS, You YS, et al. Decreased serum brain-derived neu-rotrophic factor levels in elderly korean with dementia. Psychiatry Investig. 2009;6:299–305. doi:10.4306/pi.2009.6.4.299

73. Sen S, Duman R, Sanacora G. Serum brain-derived neurotrophic factor, depression, and antidepressant medications: meta-analyses and implications. Biol Psychiatry. 2008;64:527–532. doi:10.1016/j. biopsych.2008.05.005

74. Yoshimura R, Mitoma M, Sugita A, et al. Effects of paroxetine or milnacipran on serum brain-derived neurotrophic factor in depressed patients. Prog Neuropsychopharmacol Biol Psychiatry. 2007;31:1034–1037. doi:10.1016/j.pnpbp.2007.03.001

75. Liu X, Zhang J, Sun D, Fan Y, Zhou H, Fu B. Fluoxetine enhances brain derived neurotrophic factor serum concentration and cognition in patients with vascular dementia. Curr Neurovasc Res. 2013 [epub ahead of print].

76. Cotman CW, Berchtold NC, Adlard PA, Perreau VM. Exercise and the brain. In: Mooren FC, Vӧlker K, eds. Molecular and Cellular Exercise Physiology. Champaign, IL, Human Kinetics; 2005:331–341. 77. Coelho FG, Vital TM, Stein AM, et al. Aerobic exercise increases

BDNF plasma levels in elderly with Alzheimer’s disease. J Alzheimers Dis. 2014;39:401–408. doi:10.3233/JAD-131073.

78. Rasmussen P, Brassard P, Adser H, et al. Evidence for a release of brain-derived neurotrophic factor from the brain during exercise. Exp Physiol. 2009;94:1062–1069. doi:10.1113/expphysiol.2009.048512